Each dependent variable for vehicle and driver data (i.e., derived variable) was analyzed using a 5 x 2
x 2 mixed model analysis of variance (ANOVA) model with stage (baseline, immediate treatment,
short term treatment, long term treatment, and transfer) as a within-subject variable and location
(Minnesota, Maryland) and time (daytime, nighttime) as between-subject variables. A difference was
considered statistically significant at or below a p value of .05. If the between subjects factor was not
statistically significant the analysis was then collapsed across those groups in an effort to control for
the variable and also provide the remaining analysis with increased power. Table 4-2 is a list of the
primary and follow-up comparisons and associated reporting data that are possible for each
dependent variable using the example of “0-5 mph over the speed limit” as the dependent variable.
It is also noted that paired comparisons were conducted to isolate differences within a main effect.
Due to the increased number of comparisons, all post hoc analyses (t-tests) were made using a
Bonferroni alpha correction. Post hoc analyses, when conducted for the stage comparisons, used a
series of planned comparison t-tests. These comparisons aligned with the research questions and
hypotheses. The comparisons included: